https://bugs.freedesktop.org/show_bug.cgi?id=101971
Mike Lothian changed:
What|Removed |Added
CC||m...@fireburn.co.uk
Extra validation is added to ir_validate to make sure this is
always updated to the correct numer of operands, as passes like
lower_instructions modify the instructions directly rather then
generating a new one.
---
src/compiler/glsl/glsl_to_nir.cpp | 4 +--
Also add a comment that this should only be used by the ir_reader
interface for testing purposes.
---
src/compiler/glsl/ir.cpp | 8 ++--
src/compiler/glsl/ir.h | 14 +++---
2 files changed, 9 insertions(+), 13 deletions(-)
diff --git a/src/compiler/glsl/ir.cpp
We are currently copying the name for each member dereference
but we can just share a single instance of the string provided
by the type.
This change also stops us recalculating the field index
repeatedly.
---
src/compiler/glsl/ast_array_index.cpp | 14 -
Rb
On July 30, 2017 4:08:48 PM Kenneth Graunke wrote:
It's a single atomic add, so it makes sense to inline it.
Improves performance in Piglit's drawoverhead microbenchmark's
"DrawArrays ( 1 VBO, 0 UBO, 0) w/ no state change" subtest by
0.400922% +/- 0.310389%
On 31/07/17 08:51, Ilia Mirkin wrote:
On Jul 30, 2017 5:38 PM, "Timothy Arceri" > wrote:
Also add a comment that this should only be used by the ir_reader
interface for testing purposes.
---
src/compiler/glsl/ir.cpp | 8
On Sun, Jul 30, 2017 at 5:49 PM, Chih-Wei Huang
wrote:
> 2017-07-28 22:59 GMT+08:00 Marek Olšák :
> > Hi,
> >
> > I've sent a request to revert this commit in 17.2. I'll keep it in
> > master, but I'll add a fix not to expose the new formats for GLX.
>
n some APU situations the reported visible size can be larger than
VRAM size. This properly clamps the value.
Surprisingly both CTS and spec seem to allow a heap type with size 0,
so this seemed like the easiest option to me.
Signed-off-by: Bas Nieuwenhuizen
Fixes: 4ae84efbc5c
On Jul 30, 2017 5:38 PM, "Timothy Arceri" wrote:
Also add a comment that this should only be used by the ir_reader
interface for testing purposes.
---
src/compiler/glsl/ir.cpp | 8 ++--
src/compiler/glsl/ir.h | 14 +++---
2 files changed, 9 insertions(+),
It's a single atomic add, so it makes sense to inline it.
Improves performance in Piglit's drawoverhead microbenchmark's
"DrawArrays ( 1 VBO, 0 UBO, 0) w/ no state change" subtest by
0.400922% +/- 0.310389% (n=350) on my i7-7700HQ.
---
src/mesa/drivers/dri/i965/brw_bufmgr.c | 6 --
On Sun, Jul 30, 2017 at 4:44 AM, Tomasz Figa wrote:
> On Sun, Jul 30, 2017 at 6:51 AM, Marek Olšák wrote:
> > Hi Tomasz,
> >
> > This is for 17.2 only. Not master. I have a different fix for master.
>
> Right, we don't need this in master, but according to
Hi Rob,
Sorry to bother you again.
This patch also breaks the srwast/llvmpipe on Android
since the red and blue are just swapped.
I guess nouveau has the same issue but
I'm waiting Mauro's confirmation.
Any comment?
--
Chih-Wei
Android-x86 project
http://www.android-x86.org
2017-07-28 22:59 GMT+08:00 Marek Olšák :
> Hi,
>
> I've sent a request to revert this commit in 17.2. I'll keep it in
> master, but I'll add a fix not to expose the new formats for GLX.
I thought it doesn't break GLX according to
the commit message (investigation by Chad
Looks good to me!
Reviewed-by: Tobias Klausmann
On 7/30/17 5:51 PM, Karol Herbst wrote:
> For mul(a, +-1) codegen could end up generating OP_MOV with a saturation flag
> set which we got simply ignored.
>
> Adding an assert for detecting more of such
For mul(a, +-1) codegen could end up generating OP_MOV with a saturation flag
set which we got simply ignored.
Adding an assert for detecting more of such issues.
Fixes wrongly rendered water in Hitman Absolution running under wine.
Also a few shaders in Mad Max and Alien Isolation produce such
According to section 5.8.4.5 of the 2.0 spec, the CL C version is chosen by:
1) If you have -cl-std=CL1.1+ use the version specified
2) If not, use the highest 1.x version that the device supports
Curiously, there is no valid value for -cl-std=CL1.0
Signed-off-by: Aaron Watry
We'll be using it to select the default language version soon.
Signed-off-by: Aaron Watry
Cc: Pierre Moreau
Cc: Jan Vesely
v2: (Pierre) Move changes to create_compiler_instance invocation to correct
patch to prevent
Copying the individual fields from the device when compiling/linking
will lead to an unnecessarily large number of fields getting passed
around.
Signed-off-by: Aaron Watry
Cc: Jan Vesey
---
src/gallium/state_trackers/clover/core/program.cpp | 9
Signed-off-by: Aaron Watry
CC: Jan Vesely
v2: base it on the device version
---
src/gallium/state_trackers/clover/llvm/invocation.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git
Signed-off-by: Aaron Watry
Cc: Pierre Moreau
---
src/gallium/state_trackers/clover/llvm/invocation.cpp | 5 +
1 file changed, 5 insertions(+)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp
device_version and device_clc_version are not necessarily the same for
devices that support CL 1.0, but have a 1.1 compiler and the necessary
extensions.
CC: Jan Vesey
---
src/gallium/state_trackers/clover/api/device.cpp | 2 +-
We'll need to be able to detect device version to define the appropriate
__OPENCL_VERSION__ header.
Signed-off-by: Aaron Watry
---
src/gallium/state_trackers/clover/llvm/invocation.cpp | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git
The device version is the maximum CL version that the device supports.
Eventually, this will be based on the features/extensions of the actual
device, but for now move it a bit closer to its eventual destination.
Signed-off-by: Aaron Watry
---
I've dropped the first patch of the previous series for now. I'm not
withdrawing it completely, just going to see if there's anything about
the user_ptr stuff that could have been causing the issue instead, and
if I'm using too big a hammer in this patch. If I convince myself of its
correctness,
What's here is:
Reviewed-By: Aaron Watry
That being said, are clEnqueueMapBuffer/clEnqueueMapImage still
affected or are those always done as blocking operations and my c++ is
weak?
--Aaron
On Fri, Jul 28, 2017 at 10:36 PM, Jan Vesely wrote:
>
On 31/07/17 06:29 AM, Bas Nieuwenhuizen wrote:
> n some APU situations the reported visible size can be larger than
Missing "I" at the beginning, vim fail? :)
> VRAM size. This properly clamps the value.
>
> Surprisingly both CTS and spec seem to allow a heap type with size 0,
> so this seemed
On 07/29/2017 08:21 AM, Jason Ekstrand wrote:
On Fri, Jul 28, 2017 at 10:14 PM, Tomasz Figa > wrote:
On Sat, Jul 29, 2017 at 2:06 PM, Jason Ekstrand
> wrote:
> On Fri, Jul 28, 2017
27 matches
Mail list logo